Verify that LaTeX conversion is doing something

[paperdoorknob] / paperdoorknob.py

diff --git a/paperdoorknob.py b/paperdoorknob.py
index b7e4349b98ab1d97ae2ff0ad6f50e4740bb6440f..32daf4f4d9372b07ee1e5eb801037ea2efa129ad 100644 (file)
--- a/paperdoorknob.py
+++ b/paperdoorknob.py
@@ -6,7 +6,14 @@
 
 
 from argparse import ArgumentParser
+import itertools
 import os.path
+import subprocess
+
+from typing import IO, Iterable
+
+from bs4 import BeautifulSoup
+from bs4.element import Tag
 import requests
 import requests_cache
 from xdg_base_dirs import xdg_cache_home
@@ -19,6 +26,12 @@ def command_line_parser() -> ArgumentParser:
         metavar='PATH',
         help='Where to keep the http cache (instead of %(default)s)',
         default=os.path.join(xdg_cache_home(), "paperdoorknob"))
+    parser.add_argument(
+        '--out',
+        help='The filename stem at which to write output ' +
+             '(eg: "%(default)s" produces %(default)s.text, %(default)s.pdf, etc.)',
+        default='book')
+    parser.add_argument('--pandoc', help='Location of the pandoc executable')
     parser.add_argument(
         '--timeout',
         help='How long to wait for HTTP requests, in seconds',
@@ -27,15 +40,65 @@ def command_line_parser() -> ArgumentParser:
     return parser
 
 
-def fetch(url: str, session: requests.Session, timeout: int) -> None:
+def fetch(url: str, session: requests.Session, timeout: int) -> BeautifulSoup:
     with session.get(url, timeout=timeout) as r:
         r.raise_for_status()
+        return BeautifulSoup(r.text, 'html.parser')
+
+
+def clean(html: BeautifulSoup) -> BeautifulSoup:
+    for eb in html.find_all("div", class_="post-edit-box"):
+        eb.decompose()
+    for footer in html.find_all("div", class_="post-footer"):
+        footer.decompose()
+    return html
+
+
+def replies(html: BeautifulSoup) -> Iterable[Tag]:
+    def text() -> Tag:
+        body = html.body
+        assert body
+        text = body.find_next("div", class_="post-post")
+        assert isinstance(text, Tag)
+        return text
+
+    def the_replies() -> Iterable[Tag]:
+        rs = html.find_all("div", class_="post-reply")
+        assert all(isinstance(r, Tag) for r in rs)
+        return rs
+
+    return itertools.chain([text()], the_replies())
+
+
+def html_to_tex(pandoc: str, tag: Tag) -> bytes:
+    return subprocess.run([pandoc, '--from=html', '--to=latex'],
+                          input=tag.encode(),
+                          stdout=subprocess.PIPE,
+                          check=True).stdout
+
+
+def process(
+        url: str,
+        session: requests.Session,
+        timeout: int,
+        texout: IO[bytes],
+        pandoc: str,
+) -> None:
+    html = clean(fetch(url, session, timeout))
+    for r in replies(html):
+        texout.write(html_to_tex(pandoc, r))
 
 
 def main() -> None:
     args = command_line_parser().parse_args()
     with requests_cache.CachedSession(args.cache_path, cache_control=True) as session:
-        fetch(args.url, session, args.timeout)
+        with open(args.out + '.tex', 'wb') as texout:
+            process(
+                args.url,
+                session,
+                args.timeout,
+                texout,
+                args.pandoc or 'pandoc')
 
 
 if __name__ == '__main__':